HD 2763 .A2 R38 no.TWT-N -3 SLC004484 /g-7-e' TWT- 3 THE WORLD BANK SECTORAL Li1RARY INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOrMEN MAY 1 2 1989 TRANSPORTATION, WATER AND TELECOMMUNICATIONS DEPARTMENT APPLICATION ASPECTS OF DIGITAL SWITCHING AND TRANSMISSION SYSTEMS January 15, 1981 Central Projects Staff This paper is one of a series issued by the Transportation, Water and Telecommunications Department for the information and guidance of Bank staff working in the telecommunications sector. It may not be published or quoted as representing the views of the Bank Group, and the Bank Group does not accept responsibility for its accuracy and completeness. APPLICATION ASPECTS OF DIGITAL SWITCHING AND TRANSMISSION SYSTEMS ABSTRACT Until the 1960s all telephone switching systems used electrome- chanical components and progressed from step-by-step to machine-driven to crossbar exchanges. The rapid and unprecedented advances in the 1960s and 1970s in computer technology and a phenomenal decline in costs of digital solid state electronic components led to the development and use first of the analog and then of the digital exchanges. The first digital exchange appeared in France in 1970 designed for small to medium-sized exchanges followed in 1976 by the commissioning of a series of very large transit exchanges in the USA. The success of these systems and the steady and dramatic plunge of electronic digital component costs set off a massive design and development effort by all manufacturers and leading administrations who, by 1979, were convinced of a digital future in respect of both switching and transmission systems, though differing somewhat on the best way of getting there. The digital systems have special features as well as cost and other advantages which make them specially attractive for developing countries. Because of the rapidity of the advance of digital switching and transmission systems--in combination the advantages increase--and because the digital exchanges differ so totally from earlier electromechanical systems, the information available to many developing country administra- tions is meager, despite the very large number of articles on the subject-- these articles deal with diverse issues and audiences, many of them not of direct interest or application to those in developing countries. This paper is based essentially on these articles and manufactur- ers' literature as well as on some discussions with manufacturers. The paper however specifically focuses in the developing country context on the features and advantages of these systems, the reasons for making a change to the new systems at an early date and how to implement the change. The paper is principally about digital switching systems but provides information on the also new digital transmission systems which complement the former. Prepared by: C. P. Vasudevan Telecommunications Adviser Transportation, Water and Telecommunications Department January 15, 1981 APPLICATION ASPECTS OF DIGITAL SWITCHING AND TRANSMISSION SYSTEMS By C. P. Vasudevan 1/ A. INTRODUCTION\-HISTORY AND OVERVIEW Al. A telephone network comprises switching, signaling and transmis- sion systems 2/ designed to interwork with each other. Until the mid 1970s the changes and improvement in these systems 3/ had been evolutionary and progressive. However, the appearance of digital switching systems in recent years marks a major change for future telephone development. This paper is essentially aimed at engineers and administrators who have some familiarity with telephone systems. A2. Until the middle 1960s all switching systems used electromechani- cal components namely relays, selectors, switches, etc. The first success- ful switching systems were step-by-step systems where selectors were oper- ated directly by dial control and the different selector stages were through-connected sequentially in a step-by-step manner. These systems were supplemented in the 1920s by systems using machine-driven selectors under register control and special interregister signaling with decadic pulsing. The common control crossbar systems emerged in the 1950s and took a leading role in the 1960s. These systems employed multistage switching networks according to the link principle implying a separation of the control and 1/ The author is Telecommunications Adviser, Transportation, Water and Telecommunications Department. The opinions expressed in this paper should not necessarily be taken as representing the views of the World Bank. The author acknowledges helpful comments from Messrs. Saunders, Gravell, Joshi, Ruud, Sathar and Scoffier. 2/ Transmission systems provide circuits from subscriber to exchange and between exchanges; switching systems connect the appropriate circuits to set up a telephone connection; signaling systems provide the dialed information and state of the network (free or busy) along the network (forward and backward) to, control switches to set up and break down the connections. 3/ These systems were analog; this paper is about digital systems. The difference between analog and digital signals is briefly explained in Annex 1 for persons not familiar with the terminology. -2- signaling logic from the switching network or matrix, and brought together control functions in specialized markers, registers and code senders/ receivers (common control with multifrequency interregister signaling). Each of these developments provided progressively better transmission and fewer noisy contacts, flexibility of routing and junction patterns not tied to dialed digits, simplified numbering schemes, faster signaling schemes, and greater reliability with less maintenance. These improvements satisfied essential requirements for providing an automatic nationwide telephone net- work and for increasing the efficiency of the junction and long distance circuit groups. These changes also provided for economies particularly when maintenance and capital costs were both taken into account. A3. Rapid and unprecedented technological advances were made in the 1960s and 1970s in two parallel areas, namely in the development of comput- ers and of solid state devices needed for digital technology--the two tech- nologies reinforced each other stimulating further advances in both. The principal digital circuit elements are logic (gates) and memory circuits. Starting from the 1960s technology advances made possible the provision of complete circuits, then system building blocks and later complete subsystems on one chip (paragraphs B14 and B15). The result was the ever-increasing number of active circuit elements in one chip (10 "gates" in the early 1960s to 50,000 gates in 1980) and a similar reduction of costs per circuit ele- ment ($1.00 in the early 1960s to .05 cent in 1980). Fueled by the availa- bility of these devices, the computer (and other digital systems) technology advanced likewise, leading to smaller units with greatly enhanced processing capabilities at significantly lower costs. The technology advances and cost reductions in these two areas are likely to continue at the same pace well into the 1990s. A4. The first application of these advances was the development and use of Stored Program Control (SPC) 1/ switching systems using computers for the control of the switching networks which comprised either crossbar, crosspoint or reed relay switching matrices. The SPC systems marked an important step forward in the switching system technology and benefited by the combination of unlimited access to information, fast processing of data, cheap memories and software which together created almost unlimited possi- bilities for implementing new functions. These SPC systems were refined and improved continuously until the late 1970s by incorporating the advances that were being made in computer and digital technology. The systems have found widespread application in most developed and a few advanced developing countries. A5. All the electromechanical and SPC switching systems dealt with in preceding paragraphs were of the analog type as were all transmission 1/ The- "programmability" of SPC exchanges provides a considerable degree of flexibility and versatility (paragraph G2). -3- systems until recently. Digital switching and transmission systems had been investigated and experimental systems built before the 1960s but these were not attractive or economical vis-a-vis analog systems with the type and cost of digital circuit elements available then. However, with the availability of new and attractive digital circuit elements in the 1960s and 1970s, the interest in digital transmission and switching systems revived. A6. The first digital transmission systems to be used in the telephone network were the 24-channel and 30-channel PCM (pulse code modulation) type systems introduced during the 1960s to provide additional circuits in exist- ing cable pairs. The early PCM systems were economical on the longer junc- tion routes in large metropolitan areas but as the costs progressively diminished over the years, the use of these systems became more widespread. A7. The first digital exchange in a public network was a small local exchange commissioned in France in 1970 followed some years later by several independent companies in the US. The integration achieved by use of digital exchanges with digital transmission (PCM) systems demonstrated the overall economies realizable with the new systems (beyond what was possible with either one without the other). The most frequent application at that time was at the rural and suburban exchange levels since at that time large digital local exchanges for urban areas could not compete in costs with analog exchanges. In 1976, AT&T began commissioning a series of very large 4-wire digital exchanges for long distance switching. These exchanges were found to be technically and economically attractive and demonstrated the soundness of digital group selectors in local or transit applications. The success of these local and transit exchanges and the steady and dramatic plunge of electronic component costs (particularly of LSI circuits and new types of memories) set off a massive design and development effort by all the leading manufacturers and administrations. By 1979 the swing to digital exchanges was general. In the International Switching Symposium held that year with more than 2,000 attendees from 47 countries, all administrations and manufacturers affirmed their faith in a digital future though differing somewhat on the best way and pace for getting there. A8. Currently available digital exchanges (and those under develop- ment) have benefited considerably from the experiences with SPC analog and earlier electromechanical exchanges. The digital exchanges offer a wide variety of features which were not technically and/or economically feasible in the earlier exchanges. Some of these features are particularly attrac- tive in the developing country environment where the telephone density (in terms of number of telephones per 100 of population and/or per square kilo- meter) is very low and the growth rate generally high compared with devel- oped countries. Under these conditions, the cost and other benefits of integrating digital switching and transmission systems become particularly attractive. Many developing and developed countries have already opted for digital systems, a number of them have placed orders for such systems and a few developing countries have digital exchanges in operation. -4- A9. The promise of attractive digital switching systems and of the benefits of integrating digital switching and transmission triggered a fresh interest in development of digital transmission systems using microwave, coaxial and lightwave media. A number of new designs are now competitive with analog transmission systems and more attractive systems are expected to become available in a few years. AlO. Reflecting the widespread interest in digital systems, there has been in recent years considerable discussions in several technical meetings and a large number of magazine articles some of which are referenced at Annex 3. This paper, based largely on information presented in these dis- cussions and papers, deals esentially with broad policy and implementation aspects principally of digital switching and briefly of digital transmission systems as they apply to developing countries. The coverage relates to the features, facilities, and advantages of digital exchanges vis-a-vis analog systems (Sections B and C), when to change (Section D) and how to implement and manage the transition (Section E). Section F deals with all these aspects in respect of digital transmission. Section G deals with a few topics aimed at a better understanding of software and lightwave transmis- sion as well as on a few issues relative to digital exchanges. B. SPC DIGITAL SYSTEMS--FEATURES, FACILITIES AND ADVANTAGES Bl. This section deals with the features and advantages of digital exchanges to help appreciate the reasons for the swing to digital systems in developed and developing countries. Since the exchanges currently in use in developing countries are largely of the electromechanical type, comparisons where made in this section are between digital and electromechanical exchanges. B2. Some SPC analog exchanges are currently deployed in a few develop- ing countries but the numbers are small. Though SPC analog exchanges repre- sent a major advance over electromechanical exchanges, and provide the bene- fits of SPC control (paragraphs B5 and B6), these systems do not provide some of the advantages of digital systems particularly those relating to costs. Since digital systems offer all the features of analog exchanges and more, future development and production by manufacturers is also likely to be concentrated on digital systems. For these reasons, the question of using SPC analog exchanges in developing countries can arise only for spe- cial reasons, e.g., whether an existing analog SPC exchange should be extended (or not) in future to provide additional capacity. - 5 - B3. The principal features of digital switching system 1/ result from one or more of the following: (a) the use of computer type processor or processors and software for executing the control of switching and other functions, e.g., signaling, centralized maintenance, charging, etc.; (b) the adoption of time division multiplex for the switching (or connecting) functions; (c) the capability to integrate digital switching and trans- mission (peripherals); and (d) the use of new digital LSI and memory components. Resulting from the above, digital exchanges provide important facilities to the administrations and to subscribers not technically and/or economically feasible with electromechanical systems. Even more important, digital exchanges provide lower overall costs and greater reliability than both SPC analog and electromechanical exchanges and these advantages are expected to increase further in the future. Digital exchanges use less space and are considerably easier to install than other systems. In addition, digital systems provide an easy and economical way to upgrade transmission standards (quality of speech) to current CCITT recommendations--in many developing countries, the quality of speech is currently well below these requirements. B4. This section will first deal with a short explanation followed by a listing of facilities and advantages of each of the four aspects listed at (a) to (d) in paragraph B3 above. This is followed by a similar treatment on distributed local switching with remote subscriber stages and on relia- bility. Stored Program-Control (SPC) B5. The principal advantages of SPC are due mainly to two factors: (a) the extremely low costs, high reliability and compactness of digital active circuits and memories--discussed later in this section; and (b) the flexi- bility, versatility and other features of a stored program (paragraph G2). Together they provide at reasonable costs unlimited access to information, fast processing, and almost unlimited flexibility and possibilities for implementing new functions. The significant benefits from SPC derives essentially from the following: 1/ All digital exchanges are Stored Program Controlled (SPC). Hence, the term "digital exchanges" in this paper means SPC digital exchanges. -6- - the control program and also the data can easily be modi- fied; hence improvements and operational changes can easily be introduced in a working exchange; and - the intelligence included in the program of the processor is more complex and sophisticated than in the wired ele- ments of an electromechanical exchange; this enables SPC exchanges to provide a variety of services and facilities (paragraph B6) not economically and/or technically feas- ible in electromechanical exchanges. B6. The new services and facilities made possible through SPC may be divided into the following three categories: (a) Operational facilities - many operational changes are made through software (as against wiring changes in electromechanical exchanges) and can be executed flexibly by a teleprinter from a remote location using standardized man-machine language. These include connection and disconnection of subscribers, changes of subscriber numbers and class of service, changes in routing, numbering and charging, etc.; - automatic number identification and toll ticketing of calls can be introduced easily and economically; - interfacing and signal matching with existing signaling is achieved flexibly through changes in software; - new and powerful common channel signaling systems become feasible with SPC systems; - capability to collect desired data (which can be changed through software) for traffic and service quality measure- ments on a continuous basis; - real time supervision and the ability to make changes in network through software commands from remote locations facilitate dynamic network management to cope with conges- tion, overload or temporary disruption of facilities; and - capability of the computer to store/exchange/use informa- tion on all above functions. - 7- (b) Maintenance facilities - clearing of exchange faults is easy since components are mounted on plug-in printed circuit boards; repairs consist of replacing a faulty board by a good one; faulty boards are repaired later in a central workshop; - availability of diagnostic programs to locate faults to a particuliar piece of equipment--ideally a single printed board--and remove it from serving live traffic, and print in a teleprinter relevant information for use by mainten- ance staff; and - feasibility of remote testing, fault location and diagnosis for centralized maintenance of a number of exchanges and associated live facilities. (c) New services for- sub-scribers, which cannot readily be pro- vided in exchanges without SPC, include among others: - abbreviated dialing; - priority calls; - call waiting or camp on busy; - call forwarding; and - automatic wake up alarm. (d) Use -of solid state digital circuit and memory elements in place of electromechanical parts are discussed later in this section. Digital Switching Matrix B7. Digital switching introduces into the switching technique the principles of pulse code modulation which was first designed as a means of transmission. It is based on the theorem that speech can be correctly reproduced once one gets a sample of the volume of the speech signal taken at intervals at a rate at least twice the desired bandwidth--for telephony at 8,000 samples per second. The level of the sampled speech signal is coded in the form of a sequence of 8 binary bits 8,000 times per second. The classical analog transmission of speech signals is replaced by the transmission of pulses at the rate of 64,000 times per second, the pulses being present or not present according to the value 1 or 0 of each bit. A number of such bits from several calls--usually 24 or 30 for transmission over subscriber and junction cables and higher numbers in special paired or coaxial cables, microwave, optical fiber and other facilities--are combined -8- into a single stream of pulses. The principle of digital switching consists of switching these coded pulses, instead of switching the speech through making or breaking metallic contacts, and uses digital active circuits and memory elements together with electronic crosspoints to route each incoming pulse (or a group of 8 pulses) to the designated output. B8. This transfer of pulses from input to output of the switching matrix is accomplished through use of two types of selectors. The first type of selector, generally designated "T" (like time), uses the fact that, in PCM, time is divided into 24, 30 or more intervals each allocated to a call. Switching is accomplished by delaying pulses of an interval "m" received over a PCM link in order to send them on the interval "n" over another PCM link. The second type of selector designated "S" (like space) is more similar to the classical switching which allows connection of one incoming wire to an outgoing wire. The difference is that in a classical analog switch the connection is established for the whole duration of the call, whereas for digital switching this connection is changed for each interval since each interval belongs to a different call. The arrangement of selectors and the number of simultaneous calls that can be switched at one time depends on speed at which the electronic gates (used as switches) can be switched between conducting and nonconducting states. Current systems are designed to handle 30 to 480 calls simultaneously and many of the switching matrices are made up of a combination of T and S type selectors. B9. The subscriber stage in digital local exchanges has two functions: (a) concentration in a switching matrix of many light traffic subscriber lines into a few high traffic trunks; and (b) interface with the subscriber loop. These functions are similar to those in analog exchanges except that, in digital exchanges, analog to digital and two-wire to four-wire conver- sions have to be additionally provided in the subscriber line interface. The subscriber interface facilities in digital exchanges are called BORSCHT-- an acronym for battery, overvoltage, ringing, supervision, codec, hybrid and testing. BORSCHT facilities for current designs of subscriber loops require the handling of high voltages and cannot be provided with standard computer type digital components which are essentially low voltage devices. Further, until recently codecs were also expensive. For these reasons early digital exchanges used analog concentration with metallic reed relay crosspoints, with codecs and hybrids in the trunks between the subscriber and group selector stages. However, low cost codecs using large scale integration and specially designed electronic components for providing BORSCHT facilities have become available and are now being used with a digital switching matrix for concentration in current designs of digital local exchanges. BIO. Some of the main characteristics which distinguish digital from analog switching are: - 9 - (a) digital switching is four-wire throughout;l/ hence, there is no difference between the switching networks of digital local and long distance exchanges; (b) the cost of the digital switching network is very small because, as in PCM transmission, advantages are realized from using time division multiplex; a large number of connections are handled simultaneously through the same crosspoint in rapid time sequence; (c) subscriber stage equipment can be placed remote from a digital exchange and connected via PCM links to reduce outside plant costs; these concentrators provide flexibil- ity and features not available with similar units connect- ed to analog exchanges; (d) digital switching matrices are nonblocking or essentially so; this feature eliminates the need to redistribute traf- fic, and permits "nailed down" through connections to be set up for fixed time point-to-point leased circuits with- out disrupting other traffic; (e) BORSCHT facilities (paragraph B9) for digital local exchanges in current production cost significantly more than for analog exchanges. Thus the additional cost of the subscriber stage in some current digital exchanges (not taking into account remote location possibilities or the impact of new designs) offsets to a large extent the cost advantage of the other subsystems vis-A-vis analog exchanges; and (f) the switching and transmission systems (with the exception of junction PCM systems) in the existing networks are analog and still have long operating lives. These repre- sent a large capital investment and hence it is inevitable that digital and analog plant must coexist for many years. The interfaces between digital and analog systems require analog-digital conversion and, where necessary, four-wire/ two-wire conversion; 1/ Four-wire transmission uses two separate paths for the go and return directions of speech (or other signals); two-wire transmission uses one path for both directions. Four-wire switching switches the two paths together requiring two sets of contacts for each connection while two- wire switching requires only one set. Subscriber circuits are all two- wire. Some trunk exchanges and similarly the interconnecting circuits have to be four-wire to maintain adequate speech quality. - 10 - (g) integration of digital switching and digital transmission provides a number of benefits (paragraphs Bl, B12); and (h) use of digital logic and memory components in place of electromechanical parts also provides a number of features and advantages (paragraphs B14-B16). Integrated Switching-and Transmission'(IST) 1/ Bll. Spurred by the continuing sophistication of digital circuit ele- ments and rapid decline of their costs, digital transmission systems are becoming progressively attractive for many applications even in analog environments. The current status on digital transmission systems and their application vis-a-vis analog systems are dealt with in Section F. B12. An interesting and useful feature of digital switching is that it allows the integration of switching and transmission. When a PCM link (24 or 30 channels on two cable pairs or a larger number of channels on other transmission media) is provided between analog exchanges, it is necessary to provide coding and decoding (codec) and/or multiplexing/demultiplexing equipments at 'each end. Though the costs of codecs and multiplexers are significantly less than of corresponding analog FDM (frequency division multiplex) channeling equipment, these costs are often a major part of the total costs of short haul circuits as in local junctions or toll or trunk circuits over short distances. If the PCM link is terminated at one or both ends on digital switching equipment, no codecs or multiplexers are required for the digital interconnection and savings can be made on circuit costs. Hence, digital switching becomes quite economical in a digital environment, that is in a network where there is a large percentage of digital PCM links. B13. In addition to possible cost savings, the integration of digital switching and transmission (IST) provides other advantages: (a) since all digital selectors (including those in digital local exchanges) and all digital PCM transmission systems such as for junction circuits, function on a four-wire basis, IST enables a connection from one digital local exchange to another (or from and a digital remote sub- scriber line unit) to be established on a loss-free (usually a 2-3 db for other reasons) basis even when the connection is established through one or more intermedi- ate digital tandem exchanges. This opens up two possi- bilities, namely (i) the overall attenuation can be reduced and the quality of speech improved to the new CCITT standards--existing analog networks in developing 1/ Also often referred to as Integrated Digital Network (IDN). - 11 - countries are generally quite deficient in this respect; and/or (ii) most of the permissible attenuation can be allocated largely to the subscriber loop enabling thinner conductor sizes (with consequent cost savings) to be used in the subscriber cable network than is possible with analog exchanges which are generally switched on a two-wire basis at the local and tandem levels; (b) since distances (or number of switching points) have no effect on attenuation and only a small impact on costs, there is greater freedom of choice in locating digital tandem exchanges than with analog tandem exchanges whose locations must be carefully chosen to minimize cable lengths; and (c) in digital exchanges, digital circuits are switched with- out bringing them down to voice level, thereby avoiding noise and distortion inherent in such conversion with analog exchanges and for analog transmission systems. Use of Digital Logic-and Memory Circuit-Elements B14. Digital systems are made up of digital circuits. The simplest of digital circuits is called a gate. Combinations of gates make more complex circuits, and more complex circuits provide more complicated digital func- tions. When more and more of these digital functions are combined, complex subsystems and finally full systems result. B15. The first digital circuits in the 1950s used discrete transistors or diodes and resistors to form a gate; these components were wired together by hand on printed circuit boards. In the early 1960s, combining such com- ponents into one solid state material called an integrated chip (IC) became a reality and the technology was called small-scale integration (SSI). These ICs were twice as thick as a piece of paper, as wide as a pencil lead, and provided about ten gates. Through the 1960s, the number of gates increased to over 1,000 on a single chip. First medium-scale integration (MSI) system building blocks were designed and then complete subsystems ushered in the age of large-scale integration (LSI) launched by the hand held calculator with all its circuitry on one chip (except the keyboard and display). The LSI integration pace continued through the 1970s to very large-scale integration (VLSI). It was not uncommon to put 50,000 gates, that is all the circuits for a complete microcomputer, on one chip. While the number of gates per chip increased by about 5,000 times during this period of less than 20 years, the area of the chip increased only about 25 times, that is a space reduction of 200 times per gate. Similarly, the cost per gate dropped from about $1 in the early 1960s to about 0.1 cent when VLSI chips were first manufactured (now about 0.05 cent) and is expected to - 12 - drop to 0.02 cent when production volume increases in a few years--a cost reduction of about 5,000 times. There were similar advances on memory devices. B16. The impact of the advances in digital technology is best demon- strated by what has happened to the cost of a medium scale computer. While in the early 1960s the hardware cost was about $30,000, in 1980 it was approximately $1,000 and it is projected to be about $100 by 1985. In addi- tion, the digital functions are available with high reliability, low power consumption, increased speed of operation, high accuracy and light weight features offered by digital integrated circuits. These enormous advances in the digital technology are the reason for the rapid and universal acceptance of digital switching systems by all administrations and manufacturers, and the current drive to progress to a digital future. B17. In digital systems, solid state devices (ICs, LSIs, memories, etc.) have replaced electromechanical parts of the crossbar and other elec- tromechanical exchanges (with the exception of subscriber stages in early digital systems, paragraph B9). The benefits therefrom are: - suppression of mechanical wear and consequently of the need for mechanical adjustments; - greater reliability and reduced maintenance; - extremely fast operation; - significant reduction of space required; - ease of installation; - ease of manufacture; and - benefit of the declining trend of prices of solid state devices, at a time when the future price trend for elec- tromechanical devices is upward. Remote Subscriber Line Terminals B18. The use of subscriber line units remotely from the local exchange (in the form of remote concentrators) has been of great interest to tele- phone engineers because such use increases the flexibility on the size of the exchange service area and reduces significantly the number of cable pairs required between the exchange and the remote unit. Even so, the use of such units have not proved economically attractive with conventional exchanges except for providing temporary relief. However, digital systems have the following features which make remote subscriber units extremely attractive: - 13 - (a) integration of switching and transmission enabling low cost PCM repeatered lines (without terminals) to be directly terminated at the exchange and at the remote unit providing additional pair gain advantages over what could be obtained by concentration alone; (b) feasibility of fast and powerful signaling between the exchange and the remote unit; (c) availability of inexpensive microprocessors to control switching, signaling and other functions in the remote units; and (d) maintenance and administrative features which permit testing, connection, disconnection, etc. at the remote unit to be carried out from the main exchange. B19. The first feature above leads to cost savings for the initial installation and future expansion. Because of the other three features, the remote unit is identical with subscriber units used in the main exchange, and the software implemented operations and maintenance procedures of a digital exchange provided in the main exchange also apply equally to the remote line units. There is little difference in fault diagnosis or fault removal whether the service is provided exclusively from the main exchange or from remote line units; this enables the use of remote units in digital exchanges without the costs of additional manpower needed for maintenance and upkeep of concentrators connected to electromechanical exchanges. B20. Digital remote line units are available as autonomous (local calls switched in the remote unit) or semiautonomous (local calls switched in the main exchange). While the current remote units require buildings or con- tainers, the increasing compactness of these units in the next few years is expected to permit housing them in cabinets or in large underground manhole- type accommodation. Reliability B21. An important requirement for switching systems is reliability of the highest order. Though digital exchanges are relatively new, much of the hardware and software subsystems used in digital exchanges were developed and used as part of earlier SPC analog exchanges and of digital computers. The design of digital exchanges has benefited substantially from the exper- ience with earlier systems. Many special features have been included in the design and architecture of digital exchanges to ensure high reliability and safety of operation, including the following: -14- (a) Hardware - solid state devices are intrinsically reliable with extremely low failure rates; the increasing degree of integration of a large number digital circuits in one chip has increased the mean time between failures (MTBF); in particular, the use of solid state devices eliminates a big source of faults caused by dirty contacts, incor- rect relay timing, etc., in electromechanical systems (or subsystems); - adequate redundancy is provided in the case of all units whose failure to operate properly can cause disruption of service to more than a small number of subscribers, with built-in continuous diagnostic systems to pinpoint mal- functions and disable such units from carrying live traf- fic. For this purpose, it is usual to duplicate facili- ties with load sharing or to provide a working and a standby unit with automatic changeover (that is, where n units are required, n+l units are provided) for all important and critical equipments such as processors, scanners, distributors, memory stores, switching matrices, etc.; - the monitoring and diagnostic system extends to the junc- tion, subscriber and long distance lines connected to digital exchanges, isolates the faulty outside circuits, etc., and except' in the case of subscriber lines, directs the traffic via alternative routes; and - quick replacement of faulty units through use of plug-in printed circuit boards. (b) Software - extensive testing of software programs and use of mainly standardized and well proven programs; - modular structure and functioning to limit the conse- quences of software faults and permit easy detection and location of such faults with the help of special tracing functions; and - automatic restart if implausible data or program derail- ment should be detected; this function maintains the operational reliability of the system by limiting the effects of serious faults; an informative printout is provided in case of restart furnishing basic data for making the necessary corrections. - 15 - B22. Experience with the working of the early digital exchanges con- firms the overall reliability of these exchanges after the initial debugging of software and shakedown period of about three months for hardware. The number of faults have been lower than for electromechanical exchanges. With more feedback from operational systems, the performance is expected to improve further. The few developing country administrations who have intro- duced digital exchanges have not encountered any special problems and have reported satisfaction about the operation of the new systems. C. DIGITAL SYSTEMS- RELATIVE COSTS VIS-A-VIS ANALOG SYSTEMS Cl. Perhaps the single most important aspect relating to the attrac- tiveness of a switching system, particularly in developing countries, is the issue of relative overall costs of exchanges and network through use of digital and analog switching systems. Some aspects relating to costs have been delt with as part of the general listing of advantages in Section B. In view of its importance, the cost information and analysis is consolidated in this section, incorporating the points already brought out and other new points. C2. In making economic comparisons, first off investments, incremental investments for expansion, and the (present value of) recurring expenses throughout the life of the equipment have to be considered. In addition, when comparing digital and analog systems, the pattern of network into which the exchange is to be introduced also affects the total costs. Cost compar- isons have also to take into account the cost savings and advantages of additional operational facilities, easier maintenance, reduced space, easier installation and other relevant factors. Last but not the least, future cost trends in respect of all of the above are important. C3. Cost comparisons in precise terms are, however, difficult for several reasons. Firstly, digital exchanges are relatively new and only a few manufacturers are producing such exchanges in any quantity; cost data is therefore based on estimates representing a consensus of manfacturers and administrations. Secondly, the cost savings from additional operational, maintenance and other facilities, while clearly representing. potential sav- ings in the future, may not always be realizable for some years because of the time needed for making readjustments to the labor force. Thirdly, a proper cost comparison of analog and digital exchanges should reflect the real situation in each case. For the analog exchanges, costs should be based on the continuation of the existing network design which is optimized for use of analog exchanges. For the digital exchanges, costs should be based on progressive network changes over time to exploit the special features of integrated digital switching and transmission to achieve an overall least cost solution. The plan assumptions for estimating these costs can only be very approximate. Fourthly, forecasting of future cost - 16 - trends is somewhat speculative though the general trends are quite obvious. Despite these limitations it is possible to reach meaningful conclusions by following a different approach. C4. To facilitate the presentation, the cost comparison first consid- ers the costs of a digital exchange in the worst case scenario for digital exchanges, namely of a single digital exchange installed in an existing analog network (that is, a stand alone exchange), and comparing these costs with the costs of an analog exchange under the same conditions which are the most attractive for analog. Thereafter, the relative cost advantages of the digital exchanges realizable through exploitation of the special character- istics of digital switching and transmission systems are discussed. Final- ly, the overall impact of all the factors are summed up. Stand Alone Digital Exchanges CS. The capital cost of a stand alone digital exchange comprise costs of exchange equipment, analog-digital interface, building space and instal- lation. The present value of operational, administrative and maintenance costs must also be taken into account for comparison purposes. C6. Transit (trunk and tandem) Exchanges; The main subsystems of transit exchanges comprise those for control, switching and peripheral (interfacing with junction and long distance circuits) functions. The application of digital technology and components for these subsystems pro- vide considerable cost savings relative to the cost of such subsystems with electromechanical components. Hence, the equipment cost of a digital tran- sit exchange even after adding the cost of analog-digital interface equip- ment needed for working in a total analog environment is significantly lower than the cost of a corresponding analog exchange. An estimate by a manufac- turer of digital exchanges (reference 6 at the end of the paper) places the cost of a digital transit exchange of 2,800 lines at about one-third the cost of a crossbar transit exchange in an analog environment (switching analog circuits) and at about one-fourth in a digital environment.l/ C7. Local'Exchanges: The cost of BORSCHT facilities (paragraph B9) to interface with subscriber loops is quite high in digital local exchanges in current production (paragraph B10) which are mostly of earlier designs. For 1/ The available price data from the few bids received by the World Bank borrowers indicate that the digital transit exchanges are lower in cost than crossbar exchanges though not by as much as has been stated above. Available price data from bids for digital local exchanges show that currently these prices are about the same as for crossbar or SPC analog local exchanges. These price data should, however, be treated as no more than indicative. - 17 - this reason, the equipment costs of a digital local exchange including the analog-digital interfaces are somewhat higher than similar costs for analog exchanges when both are working in an analog environment--with such costs for digital exchanges being close to those for analog exchanges when the traffic per subscriber line is high as in developing countries. However, when digital local exchanges incorporating new technology for the subscriber stages (paragraph B9) become available in a few years, the equipment costs of digital local exchanges are expected to become fully competitive with those for analog exchanges. C8. Other Costs: The installation, testing and commissioning effort for digital exchanges is significantly less than for analog exchanges because digital exchanges are usually assembled and tested as complete units in the factory before shipment. The mechanical installation and wiring effort and time, and hence their costs, are also considerably less for digital exchanges than for analog exchanges. Digital exchanges require only about 10 percent to 30 percent of the space needed for a corresponding crossbar exchange, with resultant cost savings. The operational and maintenance costs for digital exchanges are lower than those for analog exchanges (paragraph C14). C9. Overall, digital local exchanges are now cost competitive with analog exchanges, and digital transit exchanges are much more so even when both types work in an analog environment. The new designs of subscriber stages in digital local exchanges will further bring down costs when these go into production in a few years. Other factors discussed below add to the cost superiority of digital systems. Combining Exchanges C1O. Digital exchanges have two special features: - large capacity exchanges terminating a large number of subscribers and trunks/junctions as well as catering to high traffic and high number of calls are technically and economically feasible. Analog electromechanical exchanges are available only in relatively small sizes because of the limitations in the call handling capacity of common control and the economic size of group selectors; - digital exchanges whether trunk, tandem or local are four-wire throughout, while analog local and tandem exchanges operate two-wire. These features enable a single digital exchange with one control and one switching matrix to be used with resultant cost savings in many applications where more than one analog exchange would otherwise be required, e.g., - 18 - where both four-wire and two-wire switching is required (for combined local, tandem and transit applications) and/or where the capacity requirements exceed the maximum economical size of analog exchanges. Integration of Switching and Transmission C11. PCM (digital) transmission systems are finding widespread accept- ance for economical (and other) reasons to provide junctions between exist- ing or new urban analog exchanges which are more than a few kilometers apart. Similarly, digital microwave, coaxial and other systems are economi- cal in some situations even in an analog environment when the distances are modest (say, less than a few hundred kilometers). Newly emerging transmis- sion systems using optical fiber technology are expected to be economical in these and other situations. Elimination of PCM multiplexing terminals (paragraph B12) when digital circuits are terminated on digital exchanges contributes to a significant reduction of costs, particularly of the large number of junctions and short haul trunks. Remote Location of Subscriber Stage 012. As explained in paragraph B17, the subscriber stage can be placed remotely from the main digital exchange and connected to it via PCM links. Such an arrangement opens up many possibilities for significantly reducing the costs of subscriber distribution--a major cost item (usually 30 percent or more) in the costs of a national telephone system. Despite some addi- tional costs of accommodation and power supplies for the remote unit, sig- nificant economies are realizable. Future Cost Trends 013. The costs of digital exchanges are expected to reduce further in the future for the following reasons: - reduction in the cost of hardware for digital exchanges because of rapidly declining cost trend of digital com- ponents (paragraph B15 and B16); - scale economies resulting from steadily and rapidly increasing production runs due to the swing to digital systems by all administrations; - productivity improvements from the tremendous development effort now being devoted by all manufacturers; and - a large part of software costs--currently a significant part of digital exchange costs--are essentially a one time expenditure; with increased sales the unit software costs will reduce. - 19 - On the other hand, the production runs of electromechanical exchanges have already started to fall and are likely to drop further in the future because of falling demand. These systems are mature and no further productivity or technology improvements are likely particularly because of the shift of interest to digital systems. Even without the above factors the general cost trend for electromechanical systems is upward because of the inflation- ary costs of materials and labor not offset by productivity gains. For these reasons, the cost advantage of digital systems relative to analog systems is expected to increase significantly in future. Recurring Costs C14. The recurring costs of a telephone system arise from: - performance of operational changes, namely connection and disconnection of subscribers, change of subscriber num- bers and class of service, change of routing, charging, etc.; - collection of traffic data and network management; and - maintenance activities, namely location and clearance of faults. The operational and maintenance facilities provided in digital exchanges (paragraph B6) together with the elimination of the need for mechanical adjustments (required in electromechanical exchanges) make operation and maintenance of digital exchanges significantly easier, and require less per- sonnel than for electromechanical exchanges. Despite the requirement of a few higher qualified staff for locating and remedying software faults, the overall recurring costs for a digital exchange are estimated to be consider- ably less than for an electromechanical exchange. Overall Costs 015. The discussions in this section demonstrate the cost benefits realizable through adoption of digital systems. Since, in most developing countries, the existing (analog) network is still in an early stage of development and the growth rate relatively high, the costs of analog-digital conversion involved in adopting digital exchanges in developing countries is likely to be much less of a factor than in developed countries. These cost benefits are increased further by planning the network expansion to make full use of the many special characteristics and the advantages of integrat- ing switching and transmission. One estimate (reference 6 at the end of this paper) places the overall savings with integrated digital switching and transmission at 35 percent of providing the same facilities using analog electromechanical exchanges. - 20 - D. TRANSITION TO DIGITAL SYSTEMS Dl. The discussions in sections B and C above demonstrate the consid- erable cost and other advantages of using digital exchanges and the addi- tional benefits of integrating digital switching and transmission systems. Hence, the objective of any telephone administration's future plans must clearly be based on progressing speedily towards an integrated digital switching and transmission network. This section deals principally with the constraints, if any, in the developing country context, which may suggest a delay in implementing digitization or in slowing the pace of implementa- tion. In this connection, the following subjects are discussed: (a) analog-digital interfacing during transition; (b) availability of technically advanced digital exchanges; and (c) manpower and training. Analog-Digital Interface D2. A special problem in introducing digital exchanges in an existing analog network (which is not present when a new analog type exchange is added) is caused by the requirement of analog-digital conversion at each interface between the analog and digital networks--and sometimes of four- wire/two-wire conversion also. Multiple conversions from analog to digital and vice versa contribute not only to added costs but also to some technical problems related to meeting national transmission standards. Ideally, the number of analog-digital conversions and vice versa should be kept to one pair which is the minimum even in a fully digital network. Strategies for minimizing the number of conversions during the transition is both influenced by the basic strategy (dicusssed in Section E) and the pace of digitization discussed below. D3. The amount of conversion equipment required for interworking is greatest in the initial stages when the digital network is a small part of the total network and considerable traffic has to flow between the existing analog and the new digital exchanges. As the proportion of digital facili- ties increases relative to the analog, the interface requirements start to taper off so that after some years no new interfaces are required, and only some rearrangement of facilities will then be necessary. D4. Developed countries have achieved a high degree of penetration of telephone service (about 30 to 75 telephones per 100 population), spread of service to all communities and an annual growth rate of telephones of 3 per- cent to 7 percent. The presence of a fully established analog network extending to all areas of the country and a slow growth rate makes the transition to largely digital working both prolonged (stretching to about - 21 - two decades or more) and difficult. The picture in developing countries is vastly different. The penetration of telephone service in a majority of developing countries ranges from 0.1 to 5 telephones per 100 population, the service is confined to the cities, towns and a small percentage of other communities, and the growth rate of telephone connections varies generally between 9 percent and 16 percent per year--growth rate for calls is higher-- without meeting demand. Thus, the number of telephones will double in 4.5 to 8 years--and the number of calls in 3 to 5 years-and double again in a similar period thereafter. Thus in a period of 10 to 12 years, the new additions--including those for replacement of some of the old facilities-- will be three times the present network. If a large part of these additions are made up of digital switching and transmission systems, the transforma- tion to an integrated digital network with its attendant cost and other benefits will have been largely achieved. Also, with a progressively and rapidly increasing proportion of digital facilities and a diminishing pro- portion of analog facilities in the total network, the problems of inter- working and interfacing will rapidly fade out. The transitional problems of conversion to digital working are, therefore, likely to be considerably less and much easier to manage in developing countries than in developed countries. Availability-of'Digital Exchanges D5. As of the end of 1980, a small number of leading manufacturers were offering proven digital exchanges; a number of such exchanges have been commissioned for regular service with telephone utilities and are function- ing satisfactorily. A number of others, including most major manufacturers, are now offering digital systems which have been tested in a few local installations but are as yet not in widespread operation with telephone utilities. Due to the continuing rapid development of logic and memory elements in the form of VLSI, bubble memory, etc., the new systems incorpor- ate some advances not available in earlier systems. This is particularly so in respect of subscriber stages of digital local exchanges (paragraph B9). D6. The issue then for developing countries is whether they should or should not continue for a few more years with the electromechanical systems until the choice for fully proven digital systems becomes wider and new designs with the improved components become available. General wisdom would lie in opting for the safe course of accepting only systems fully proven in public use. Accordingly, the decision has to be based on whether the cur- rently available proven digital systems are sufficiently good and there is adequate choice for selection of a new system vis-a-vis the disadvantages of continuing with electromechanical systems for a few more years. In addi- tion, it may also be appropriate to weigh the benefits and risks of going in for a new system not fully proven in widespread use. - 22 - D7. As of the end of 1980, fully proven digital systems of either the CEPT (32-channel PCM) or the North American (24-channel PCM) compatible types are available currently from at least three manufacturers from three different countries for each type. In practice, the choice has not beer significantly greater for crossbar exchanges. Even in the initial stages, these digital systems have performed at least as well if not better thar conventional systems. D8. As far as technology is concerned, it is and will be a continuous development process similar to what has been happening in the transmissior field with, say, microwave systems. Currently available digital switching systems have benefited from computer technology and earlier SPC analog experiences, with the result that they incorporate advanced (though not the latest) architectural and design features. These systems will change ir future with continuous redesign of subsystems using the technical advances in such a way that the new digital exchange designs will be fully compatible with existing digital exchanges on a modular or a plug-in replacement (or add on) basis. Maximum changes in the next decade in digital switching systems are likely to be in the subscriber stage subsystem and in software. The currently available digital exchanges will not, therefore, become obsolete or incompatible with new designs. D9. Thus it would seem that there are few substantive reasons for postponement of a decision to digitalize the switching systems even where a telephone administration decides to stick to only fully proven systems. However, if an administration is willing to accept digital systems which are not extensively field proven, the choice of systems (including designs of Z later generation) even now becomes wider and the competition increased. The price for these benefits is additional risk. The extent of the risk anc ways to minimize them are discussed below. D1O. Experience with new computer models, and several types and designs of SPC analog and digital exchanges, has shown clearly that there are few, if any, problems in the hardware part of the new systems. In general, after the break-in period of a few months, the performance and reliability of the hardware are, if anything, superior to older systems. The principal prob- lems in introducing new systems are in the software, most of which are usually spotted and resolved during installation and commissioning tests anc in the first few months of operation. The resolution of the few residual problems require the commitment, competence, and cooperation of the manufac- turer for prompt action. Where these are assured, the necessary changes are easily implemented because these are in software--unlike changes to compon- ents and wiring for overcoming similar problems in electromechanical exchanges. Further, many of the interface problems encountered on introduc- tion of new electromechanical exchanges into a network were caused by rela3 timing problems and should not occur with digital systems, since these dc not have relays and respond instantaneously to signal changes. In addition, any remedial measures required for signal matching are also in software anc easily incorporated. - 23 - Dl. For these reasons, the principal risk of accepting digital systems not fully field proven is most likely to be some delay in the commissioning of the early exchanges, provided the supplier has the commitment and the competence to remove troubles as they are encountered. Further, many of the newer digital exchanges are already in limited operation and additional experience will have been gained before their delivery on orders placed hereafter. D12. In the case of manufacturers with a long history and experience of satisfactory international operations and with substantial investment, effort and funds expended in development of digital switching systems, the risks other than some delays are not likely to be significant. It is on this basis that a large number of telephone administrations have ordered systems even before they have been fully tested in the field. Their exper- ience so far has been satisfactory. The decision on whether to accept digital systems other than fully proven is for an administration to make taking into account all the available data. Manpower and Training D13. An important prerequisite for the successful implementation of digital exchanges is the availability of trained manpower to plan and install these exchanges and to handle the operational and maintenance tasks. In general, the existing staff can be easily retrained to handle most of these tasks, but new staff with qualifications in computer program- ing will be required to handle the software aspects. D14. Planning of digital systems is no different from planning for analog exchanges and networks and the existing planning organization should be able to handle the new tasks after an orientation course to familiarize the staff with the roles of software and hardware in digital exchanges and their special features. However, additional staff trained in computer tech- nology will be needed to do the site application software (paragraph G14) which corresponds to engineering of electromechanical exchanges. D15. Installation of digital exchanges is similar to transmission systems and requires only a fraction of the manpower required for electrome- chanical exchanges. Because of the modularity of the digital systems, con- struction with plug-in printed circuits, compactness and use of software for meeting individual requirements, a considerable amount of initial testing and debugging of digital exchanges are carried out in the factory before shipment. Even so, installation and commissioning tests have to be redone at site. Most of these tasks are greatly aided by the availability of com- puter based test equipment with custom software which goes through the rou- tines and helps localize troubles. Existing installation staff can handle most of these tasks after a few weeks of training. However, software fault diagnosis and removal will require a small number of new staff with software skills. Hardware faults are remedied by replacement of the faulty units, e.g., printed circuit cards. Since no electromechanical adjustments are - 24 - needed, the staff required for testing and commissioning digital exchanges is small relative to the staff needed for electromechanical digital exchanges. Even so, it may be desirable to entrust the responsibility for installation and commissioning of at least the first few digital exchanges to the supplier, while associating local staff with the work to get on-the- job training and experience. D16. The staff required for maintenance and operation of digital exchanges is less than for conventional exchanges. Most of the operation and maintenance tasks can be satisfactorily carried out by existing staff after some retraining. The exchange administration, namely the tasks of connecting and disconnecting subscribers and effecting traffic/routing changes (which in conventional systems require wiring changes by semiskilled technicians) are carried out in digital exchanges through software changes from a remote teleprinter and can be done by nontechnical staff after a few weeks' training. Similarly, the existing technicians can be retrained in a few weeks to understand the routine use of diagnostic and fault location features of digital exchanges and for replacement of faulty cards. The repair of faulty cards can be done locally in a central workshop and would require a few technicians with skills in electronic circuit testing repair-- an alternative is to send them to the supplier for repair and return. In addition, a few specialized software staff are required to attend to sof- tware troubles of other than a routine nature. The training (or retraining) of initial numbers of staff to perform the new tasks should be made the responsibility of the supplier. In addition, local training facilities in the training school should also be established with assistance from the sup- plier personnel. D17. In some ways, the staffing for digital exchanges is less of a problem than for conventional exchanges since the numbers required are less and no mechanical skills for adjustment of relays and switches are called for. However, recruitment and training of a few specialized software quali- fied persons for complex troubleshooting, and for preparing site application software may be a problem in some developing countries. Where qualified software personnel for the complex tasks are not available locally, a few expatriate specialists may have to be retained for some years. Other tasks can be handled by retraining existing nontechnical and technician staff. Summing Up D18. It is obvious that, in developing countries, adoption and use of digital switching and transmission systems should not be delayed. Any addi- tion of analog exchange capacity hereafter would imply foregoing of the benefits of digital switching and increase the need for and cost of analog- digital conversion equipment to interface with a future digital network. Hence, from now on, expansion plans in developing countries for all exchanges and transmission facilities should generally favor digital sys- tems, and use of analog systems should occur only where they are justified economically or otherwise after taking into account all the factors. - 25 - E. IMPLEMENTATION OF DIGITIZATION El. The existing networks in all countries, extensive and fully devel- oped in industrial countries and in the early stages of development in developing countries, use analog switching and mostly analog transmission facilities, many of which have a residual economic life. Since it is inev- itable that analog and digital systems must coexist for many years, the new digital facilities must interwork with an existing analog network. An acceptable strategy for implementing an integrated digital switching and transmission network should aim at maximizing the economic and performance benefits in the long run, while minimizing the short-run economic and tech- nical penalties, mainly because of having to provide more than the minimum of analog-digital conversion (paragraph D2). It is presumed that additional analog capacities will be created only when the economic and technical advantages outweigh the disadvantages. E2. This section deals first with the implementation strategy during transition to full digitization. This is followed by the overall structure of the ultimate digital network to serve as a framework to guide detailed planning during transition. Finally, a few topics related to the transition are discussed. A Strategy for Digitization E3. The following strategy options (among others) are generally put forward for introducing digital technology: (a) digital equipment is used wherever a new need arises for switching equipment and circuits, or obsolete equipment requires replacement for technical or economic reasons; (b) digital islands: old equipment is replaced completely with digital equipment in limited geographic areas; all the features of the new networks are available in these "islands"; replaced equipment can be used to provide extensions to other exchanges of similar types; (c) overlay digital network: a new digital network is superimposed upon the old network and all extensions of the network are made with digital equipment. Subscrib- ers connected to the old network and demanding new ser- vices are transferred to digital exchanges. The old part of the network is replaced gradually until finally scrapped. Interconnections are provided between the new and old networks at suitable places. In practice, the introduction of digital systems will not proceed exclusive- ly along any one of the above options but by a combination of perhaps all of the above depending on the circumstances in each application. - 26 - E4. Option (a) in the above paragraph is logical for rapid digitiza- tion and will also be the most ecnomical if digital transmission links can also be justified and provided at the same time to interconnect the digital exchanges. If the latter is not possible for any reason, the traffic between digital exchanges will have to be routed via analog switching/trans- mission facilities with some economic and technical penalties. Option (c) ensures that all traffic between digital exchanges is routed via digital facilities but may require some digital facilities to be created ahead of time. E5. In developing countries, the existing network is generally in an early stage of development and the growth rate is high (paragraph D4). Further, it is usual to find that the installed capacities of existing switching and transmission facilities are either inadequate to meet demand or are fully used up with little or no spare capacity left. Hence, the requirement of meeting demand requires expansion of both switching and transmission facilities all over the country at a rapid rate. E6. Since digital facilities are cost competitive for most new instal- lations, a considerable amount of digital switching and transmission facili- ties will be created under option (a) in a relatively short period. While this does not assure that new digital transmission media will be available in all cases to interconnect digital exchanges when they are commissioned, this can be achieved by making appropriate adjustments to the program by scheduling the works with an eye on achieving a reasonable degree of inter- connection within the digital network, namely combining the best features of option (a) and option (c). Even so, some digital exchanges may have to be temporarily interconnected using analog facilities for economic or other reasons. Analog-digital conversion equipment used for such interconnections can be redeployed at a later date when digital facilities become available. An appropriate strategy for digitization in a developing country should be exactly the above. Option (b) is likely to be appropriate when service is to be extended to new rural communities--for this application, an appro- priate approach is likely to be to install a new digital exchange located suitably to replace one or more electromechanical exchanges (to be rede- ployed elsewhere) and provide service to rural communities through use of remote line units and digital transmission systems, e.g., digital UHF radio links. Overall Network Structure E7. The following features of an all-digital network are relevant to the structure of a national routing plan: (a) all digital exchanges are SPC controlled with full flexi- bility of routing; (b) wide range of economic sizes from small to very large units; - 27 - (c) full flexibility to combine long distance (trunk), tandem and local exchanges in one unit; (d) availability of near ideal remote subscriber line units; (e) digital trunk and tandem exchanges cost considerably less than corresponding analog exchanges; and (f) since digital PCM equipment is built in multiples of 30 (or 24) channels, the minimum number of channels in a circuit group (and therefore its efficiency) is likely to be higher than in an analog network. E8. In view of the above, an economical structure for the digital net- work will have the following features as compared with the present analog structure: (a) the network will be based on a hierarchical structure sim- ilar to the present, but the number of hierarchical levels will decrease; (b) the number of digital trunk exchanges will be fewer than in an analog network and their sizes larger; (c) these trunk exchanges will be interconnected by large capacity trunk groups, thus offering benefits of high efficiency; however, for reasons of security, extensive meshing of exchanges is also desirable and PCM high usage first choice routes will be provided where there is sufficient community of interest; and (d) new local switching units and remote line units will be located largely where earlier switching units were placed or at distribution points in the present cable network, since these reflect the community of interest. Flexibility of Digital Exchanges E9. Digital exchanges have a number of features which greatly facili- tate the planning and implementation of the transition to digitization. These include some features mentioned briefly earlier--these are treated more extensively below--and others not previously mentioned. These features provide flexibility from a combination of the following: (a) with the architecture used in most digital exchanges, e.g., multiple processors, distributed controls, modular switching matrices, etc., a trunk, tandem or local exchange installation (or even a combination of these in one installation) can be expanded without economic - 28 - penalties from small sizes of 1,000 or 2,000 (equivalent local) lines to very large sizes of 100,000 or more equiv- alent lines; the modular functional design and physical construction together with the flexibility of software enable extensions of installations to be carried out with relative ease and minimum disruption to working exchanges; (b) low cost of PCM repeatered lines to interconnect exchanges, and distributed switching (viz. ability to locate subscriber stages away from an exchange) provides great flexibility concerning location, service area and size of exchanges; and for rearrangement later, it if becomes necessary; (c) low costs and sophisticated control facilities of digital exchanges add to the flexibility of routing; nonblocking features of the switching motion in digital exchanges sim- plifies traffic engineering; (d) signaling interfaces with existing exchanges is achieved through software; and (e) through use of digital exchanges large increases in capac- ity are possible in existing buildings through more effic- ient use of available space and/or progressive replacement of electromechanical exchanges for technical or economic reasons; new buildings where needed can be located in modest sized sites; these eliminate a common bottleneck holding up expansion. Replacement and Extension of Analog Exchanges ElO. Replacement: Existing analog exchanges represent a considerable investment. Replacement of such exchanges requires additional measures (hard to come by in developing countries) without adding to capacity. Hence, replacement of existing exchanges will ordinarily be hard to justify except where such exchanges have outlived their usefulness. The following considerations are relevant to this issue: (a) standard of service is below acceptable limits and/or maintence effort is unduly high because of deterioration with age; (b) manufacture of the particular type of exchange has been discontinued so that interfacing relay sets are difficult to obtain; (c) cost of spare parts is high; - 29 - (d) existing exchange cannot carry increasing traffic and can- not be extended because of design limitations; and (e) space vacated by replacement is needed for providing addi- tional capacity possible with digital exchanges. Ell. Extension: Extension of an existing analog exchange to provide additional capacity should be considered only if this approach is clearly favorable on technical and economic grounds as compared with a digital alternative, namely a new digital exchange or a remote subscriber line unit. Since a remote unit is in effect an extension of a nearby digital exchange (which may be a trunk, tandem or local exchange), extensions to existing analog exchanges would not generally be advantageous. This is par- ticularly so for trunk or tandem exchanges where the technical and economic advantages of digital exchanges are considerable as compared with their analog counterparts. Installation of additional analog capacity may be justified if the extensions are to relatively modern equipment, with long residual life, are small relative to the installed capacity, and future spare parts availability is assured. Planning for Digitization E12. The existing routine plans were designed on the basis of analog systems; routine plans for digital systems would be different (paragraph E8). This would suggest the design of new routing plans at the national level and for each large city network. However, digital exchanges permit considerable freedom as compared with analog exchanges in respect of location, size, ease of expansion, routing capabilities in transit and local exchanges, traffic, etc. Moreover, the new digital exchanges or remote line units will be located at (or near) the existing exchanges or at the cable distribution cabinets since these represent the community of interest and enable the existing routes to be exploited economically for providing additional cir- cuits. Hence the main decisions in a revised routing plan will be based on which of the present transit and local exchanges will continue as such and on the number and size of the remote line units to minimize cable distribu- tion costs. Most such decisions are likely to be obvious and the remaining ones can be resolved by simple studies as and when required. Hence the implementation of digitization without a comprehensive revision of the fundamental routing plans is not likely to cause economic penalties or other difficulties--a series of pragmatic short-term plans should be adequate in the developing country context where for the most part the national network is yet to reach the stage of development where all the hierarchical levels are in place. - 30 - F. DIGITAL'TRANSMISSION SYSTEMS Fl. Digital transmission systems have a number of advantages over analog transmission systems which are enhanced when used with digital exchanges--the justification for the latter is partly on this account. Hence a paper on digital exchanges is incomplete without a brief discussion of digital transmission systems. F2. The application of digital transmission systems has so far been mainly in the industrial countries and a few developing countries, largely for providing additional circuits on existing junction cables. The other digital transmission systems are still in a stage of development though these have advanced to the point of large scale application in a few indus- trial countries. This section deals briefly with all the aspects relating to these systems as has been covered in sections B to E for digital exchanges. Advantages of Digital Transmission F3. Compared with analog transmission, digital transmission has the following advantages: (a) transmission quality is independent of distance since digital signals are regenerated at intermediate points without loss of quality--in analog systems not only the signal but the noise is also amplified at intermediate repeaters; (b) a PCM link being a digital medium is specially suited for efficient transmission of digital messages, e.g., data, teletext, facsimile, viewdata, telex, etc; a PCM voice channel has a capacity of 64 kilobits per second which makes it a very powerful data channel (analog voice cir- cuits of good quality transmit data at 9.6 kilobits/ second); (c) time division multiplex allows an increase in capacity on cable pairs originally used for single telephone chan- nels; PCM transmission can economically provide addition- al circuits in these cables; (d) new wide band transmission media such as waveguides and optical fibers are essentially digital media; (e) digital multiplexing terminals are considerably less expensive than similar analog multiplexing equipment; overall digital transmission equipment is more economical than analog systems over a range from about a few kilo- meters to a few hundred kilometers; when used with - 31 - digital exchanges at one or both ends, digital transmis- sion is attractive from zero to about 900 kilometers; and (f) future cost and reliability trends of digital integrated chips are likely to increase the attractiveness of digital transmission; Other advantages of integrated transmission and switching are given in para- graphs B12 and B13. Introduction to Digital Transmission Systems F4. Digital transmission systems for use by telephone administrations are based on the use of Pulse Code Modulation (PCM). These are built up of the digital multiplexing equipment or the PCM terminal and the PCM transmis- sion line. The multiphase equipment converts a number (30 or 24) of analog signals to a digital signal on the transmitting side and the inverse func- tions on the receiving side. The transmission line which can use different media, e.g., paired cables, microwave, coaxial, optic fiber, etc., conveys the digital signals between the two multiplex terminals. PCM Terminal Equipment F5. As in analog systems, the terminal multiplexing equipment is based on the use of a primary or first order multiplex equipment which converts 30 (or 24) analog speech channels together with associated signaling into a digital signal. These first order digital signals are combined (similar to combining of 12-channel groups to 60-channel groups, etc.) to form the basis for hierarchies of digital multiplexing terminal equipment. F6. CCITT has recommended two different standards for PCM systems--one based on CEPT (European) and the other on AT&T (American) standards. Both the CEPT and American type systems use sampling at 8,000 times per second and 8 bits per sample. Other features, using number of channels, speech coding and signaling arrangements are quite different. The two types are not compatible. F7. In the CEPT system, the primary or first order multiplexing equip- ment, contains 32 time slots of which 30 are voice channel time slots, one time slot for associated signaling and one time slot for link synchroniza- tion. The American system contains 24 voice channel time slots and no spe- cial time slot is assigned for signaling or synchronization. The channel associated signaling is achieved by taking the least significant bit in every PCM channel every sixth frame for signaling purposes. For synchroni- zation, one extra bit is inserted in the frame. This bit can also be used for common channel signaling. A summary of important technical data for the two standards of multiplexes is given in Annex 2. A fuller description of PCM and of the digital signal multiplex is given in reference 9 listed at Annex 3. Factors affecting the choice between the two types are discussed in paragraphs G25-G26. - 32 - PCM Systems for Cables in-Local Networks F8. PCM 30-channel (or 24-channel) systems operating over metallic cable pairs used for junctions (or for subscriber distribution between the exchange and remote line units) is the most widespread application of PCM systems. The systems use two pairs, one for each direction. The repeaters are buried underground in loading pot type cases. For 30-channel systems, the distance between repeaters can be up to 1.8 km for 0.6 conductor pairs and up to 3.2 km for 0.9 mm pairs. Power is fed from the line terminating repeater at the exchange for distances currently up to 13 km per 0.6 mm pairs and up to 22 km per 0.9 mm pairs are now available--larger distances can be powered with larger conductor sizes (say, up to 70 km with 1.4 mm pairs). Full supervision, fault location and diagnosis are carried out from the terminal station. F9. On existing junction and subscriber cables designed for use at voice frequencies, only a fraction of the pairs in a cable can be used for PCM working because of limitations of crosstalk. Even so, it is possible to increase the circuit capacity of these cables about sevenfold through use of PCM. Recently cables specially designed for enabling PCM systems to be applied on all pairs have become available. These use polythene insulation, pairs of lower mutual capacities than in normal junction and subscriber dis- tributon cables, and half the pairs in each cable (for the go and return directions) are separately enclosed in aluminum screens. All suitable pairs in existing cabes will first be exploited but new routes will perhaps be equipped with cables of the new type where these would be cost competitive with optical fiber cables. FlO. When PCM junctions are to terminate on digital exchanges at both ends, no terminal multiplexing equipment is required and the transmission line equipment is directly connected to the exchange. Under these condi- tions, PCM transmission is always less expensive than analog junctions on cable pairs, even when spare cable pairs are available because the analog digital conversion equipment at both ends is more expensive than converting the line for PCM working. Fll. When the junctions terminate on a digital exchange at one end and an analog exchange at the other, analog cable pairs are less expensive than PCM working if the distance is less than say 3 kilometers. Even so, there are many advantages in using PCM systems even over these short distances: (a) less attenuation since PCM systems can be operated on a loss free basis; (b) lower conversion costs in the long run because the PCM equipment can be redeployed later when the analog exchange changes to digital in future; (c) no distance constraints on signaling; and - 33 - (d) signal matching can be done by locating relay sets at the analog exchanges and making software adaptation at the digital exchange. The relay sets at the analog exchanges are of the same type as is used for PCM links between analog exchanges. Digital Radio Systems F12. Radio systems using microwave and UHF frequencies 1/ are providing and will continue to provide a large part of the long distance circuits. The radio systems now in use are of the analog type. The interest in digital radio systems is relatively recent (following the swing to digital exchanges) and the digital transmission systems development is still in the early stages. Currently available digital radio systems have some limita- tions vis-a-vis analog systems, e.g., smaller number of channels per radio bearer for a given frequency spacing (but not necessarily the total number of circuits, since frequencies are more easily reused in digital systems than in analog systems), shorter hop distances, and need for higher power or diversity arrangements relative to analog systems. Some recent designs have somewhat reduced these shortcomings, and further improvements are likely in the coming years. F13. The first digital radio systems developed and applied were in the 11, 13 and 15 GHZ bands, providing high capacity PCM links for use as urban junctions. These systems are particularly attractive when operating between digital exchanges and where laying of cables is difficult. F14. Many major administrations are now commited to establishing digital transmission facilites. A number of new radio links in the 2, 7 and 8 GHZ are already available and systems in other frequency bands are under design. These systems have channel capacities which are from one-half to three-fourths of the channel capacity per radio bearer of analog systems in the same bands and frequency allocation. These systems can use the existing infrastructure (towers, antennas, feeders, etc.) to provide additional radio beams with digital terminals as repeaters and, if necessary, sharing these facilities with analog radio beams. F15. Digital microwave repeaters are currently somewhat more expensive than analog repeaters. These costs are offset by the lower costs of digital multiplexing equipment when the circuits are to be brought down to voice level--no terminal multiplexing equipment is needed when terminating on digital exchanges. Digital radio systems specially for low to medium 1/ The terms microwave and UHF used here are different from CCITT classifi- cation for these terms. Here, microwave systems are those operating on frequencies above 1 GHZ and UHF systems those operating on the 400-900 MHZ frequencies. -34 - capacity routes are now likely to be cost competitive with analog systems for distances up to say 600 to 900 km, the latter when terminating on digital exchanges. Fl6. Many manufacturers also offer a digital insertion unit for FM (analog) radio systems which allows additional transmission of a 2 M bit/sec signal (30 channels) above the base band of 960 or 1,800 channels. Another unit now becoming available is a "transmuxer" or "transmultiplexer" which directly converts a 60-channel supergroup into two 30-channel PCM line interfaces and vice versa, using modern digital technology and costing less than is possible with conventional analog digital conversion at voice level. (The American arrangement is to convert two 12 channel analog groups to one 24-channel PCM.) F17. Where an analog microwave system exists, the following options are available for terminating circuits from these sytems on digital exchanges: (a) leaving the FM (analog) microwave system without change and either using analog digital conversion at the indi- vidual channel (voice) level or using transmuxers at the supergroup level; (b) adding digital radio bearers to the system, where analog systems continue to have a useful application, e.g., pro- viding very long haul circuits or circuits to be termin- ated on analog exchanges; (c) conversion of the whole route to digital working using the existing towers, power plant, etc. The new system will have a reduced ultimate capacity; and (d) adding a digital insertion unit to work above the analog baseband. This can provide only 30 channels per radio bearer. Digital Coaxial'Cable Systems F18. Coaxial cable systems are mainly used in developed countries but are also used in some of the larger developing countries. Microwave systems provide the large part of long distance circuits in developing countries. A number of digital transmission systems for use on coaxial cables are now available. Compared to analog systems, these digital systems currently pro- vide 30 to 50 percent less number of channels per pair of coaxial tubes (with the same repeater spacing), but new digital systems particularly of large capacity are under development and, when available, will provide more numbers of channels per coaxial pair than the analog systems. Currently, coaxial digital systems are cost competitive with analog systems only for short to medium distances (up to a few hundred kilometers) where the lower costs of PCM multiplexing equipment, or zero cost when terminating on - 35 - digital exchanges, offset the higher costs of PCM repeaters. In future, digital coaxial systems are likely to become cost competitive with FDM (analog) coaxial systems, but by 1985 or so, optical fiber cable (lightwave) systems are likely to be less expensive than, and provide other advantages over, coaxial cable systems particularly for large route capacities. Even then, digital coaxial systems will continue to have useful applications for providing digital communication over spare coaxial tubes on existing routes, or for replacing present analog systems by digital systems. Optical Fiber Cable*(Lightwave) Systems F19. Not since microwave has there been as significant a transmission technology development in telecommunications as the lightwave technology which already shows great promise for providing a wide variety of circuits-- from subscriber loops to long haul applications. At present, several exper- imental installations of high capacity/long distance digital transmission systems using optical fiber are working in several countries and others are being developed in close cooperation between administration and industry. Based on the satisfactory experience from these early installations, some of the major countries are embarking on establishing high capacity optical fiber cable systems for regular commercial use--e.g., the Boston-New York- Philadelphia-Washington route to be commissioned in the next few years in the USA. F20. Optical fibers have several advantages over conventional paired cables and coaxial cables. They have low transmission loss (with long repeater spacings), large band width (large capacity for circuits), small cable size and weight, and immunity from electromagnetic interference (no lightning or power line induced high voltage conditions and no crosstalk problems). The consensus is that these systems are extremely reliable with long MTBFs (mean time between failures) and require little maintenance; also, the installation of the optic fiber cables can be handled by current cable jointers with only a few weeks training. F21. At present, the lightwave technology has advanced to the point at which there are no foreseeable technical problems which would prevent the use of lightwave systems in any segment of telecommunications, viz. loops, terrestrial long-haul, undersea or on premises. The barriers to application are purely economic. Optical fiber systems are now cost competitive with coaxial cables for high capacity long distance routes. The range of appli- cation for these systems are expected to increase rapidly with increases in production volume (and resulting increases in scale economies), new techno- logical advances, the emergence of digital switching (and therefore of IST) and the increases in cost of alternative systems. The next large scale use of lightwave systems can be expected in large metropolitan areas for large capacity junction routes between digital exchanges where, because of the larger repeater spacing, the repeaters can mostly be housed in exchange buildings. In a decade or so, lightwave systems are expected to be cost competitive even for routes providing modest capacities. - 36 - F22. For developing countries, the lightwave systems have only a limited role at present because existing routes can be expanded with PCM systems at a small incremental cost, but the potential for future applica- tions is very great. Lightwave systems developments are expected to be mostly related to digital transmission because this medium is particularly suited to this mode. Some technical information on lightwave systems is given in Section G. G. SELECTED TOPICS Gl. This section deals with the following topics related to digital switching and transmission systems: (a) software; (b) technical aspects of lightwave communication; (c) choice between 24-channel and 30-channel PCM systems; (d) manufacturer support needs; (e) procurement; and (f) forthcoming CCITT publications on digital systems. Software G2. The flexibility and versatility of the digital computer is derived from its programmable nature. The computer performs exactly the operations it is instructed to do. The list of instructions to the computer to achieve the desired result is called the 'program.' With an appropriate program stored in the machine and using basically the same hardware, a stored pro- gram control (SPC) telephone exchange can do, exactly like a computer, any desired task. By changing the program, the behavior of the exchange can be modified or changed. G3. Any instruction given to a digital processor (computer) in a tele- phone exchange must be in digital code of the machine so that it can sense the code, decide what instruction the code conveys and act to execute the instruction. This digital code which the machine understands is called machine code. This code is composed of 'O's and 'l's that represent the numbers, letters, symbols, commands, etc., issued to the machine. When rhe detailed and step-by-step instructions for the computer are programed directly in such a code, then the program is being written in machine lang- uage, and is called a machine language program. G4. On the other hand, human communication is different like telling a taxi to take you to the airport. The taxi driver hears and understands the - 37 - instruction and does what is required. A computer does not directly under- stand the human language. G5. The conversion of instruction in the human language to the machine language if done by humans is tedious, detailed and slow, with possibilities of errors and difficulties in locating and correcting the errors. Usually this task is done by a computer which is good for this purpose. The program that the computer follows is called an Assembler to move the conversion to machine code one step closer to the human language. G6. Humans can instruct the processor by selecting a mnemonic that is an abbreviation of what the machine does. The programming with mnemonics is called assembly language programing because, after the sequence is written, it is fed into the assembler program which makes the conversion to machine code and arranges it into memory in proper order. G7. Even these programs (with mnemonics) are not like the normal lang- uage that humans speak and understand. Writing programs in a language closer to human language is called high-level language programing. Another type of computer program is required to convert from high-level language statements to the machine code. This is called the compiler and is more involved since the conversion is much more difficult. The easier the pro- graming is made by bringing the language closer to the human language, the more complex the computer program needed to convert the machine language statements in machine code. But once the conversion program is available, it can be used over and over again as necessary. G8. This conversion process (programing language to machine language) using a compiler and/or assembler would typically be performed using an "off line" data processing system. After the conversion process, a "load" tape in machine language would then be available for use in the SPC switching equipment. G9. A number of computer languages have been devised to suit particu- lar applications. Each language has its own terms, grammar and syntax. The programs have to strictly follow the language requirements for the processor to interpret and carry out the instructions. G10. Software comprises a set of computer programs, procedures, and possibly documentation concerned with the operation of the processor, e.g., compilers, library routines, manuals, circuit diagrams, etc. Gll. Programs for SPC Exchanges: CCITT has recently drawn up specifi- cations for a high-level language named CHILL (CCITT high-level programing language) and two other related documents. CCITT has recommended that manu- facturers adopt this language for SPC telephone exchanges--and many manufac- turers are proposing to do so. The concept of having a standardized high- level language will permit a telephone administration to develop and main- tain software without providing an extensive training for each type of equipment. - 38 - Gl2. Programs for SPC exchanges can be broadly divided into two cate- gories, viz. generic program, and site application program or data base. G13. Generic Program: A generic program refers to that part of the SPC software which is common to a specific class or family of switching equip- ment. For example, the reception and registration of digits by an incoming trunk would be performed using the same program or subroutine, that is gen- eric, without regard to the specific office location or network application. G14. Site Application Program or Data Base: Each individual telephone exchange has a series of specific characteristics that are not common to any other location, e.g., subscriber directory numbers, line and trunk termina- tions, routing pattern, trunk group signaling and group size (number of individual trunks), office hardware configuration, etc. These constitute the data base for the station. During the operation of the SPC equipment (call processing), the generic programs utilize the data base to determine what actions are to be taken, appropriate to the subscriber's class of ser- vice, digits dialed, etc. The data base has to be compiled when a new exchange is to be commissioned and is altered for connecting or disconnect- ing subscribers, changes in routing, signaling, etc. Technical Aspects,of Lightwave (Fiber Optic) Communication Systems G15. In lightwave systems, the originating electrical signal is con- verted to a light signal by a light source, e.g., a light emitting diode or laser. The light source couples the light into a glass fiber for transmis- sion. Periodically along the fiber, the light signal is regenerated by a lightwave repeater unit. At its destination, the light is sensed by a spe- cial receiver and converted back to an electrical signal after which it is processed conventionally. G16. The first communication system based on use of optic fiber as a transmission medium was demonstrated on a benchtop in the United Kingdom in 1970--ten years ago. The telecommunication community was greatly excited with the prospects such a system could offer and convinced enough to invest substantial manpower and capital in the next decade to develop this technol- ogy. A number of experimental links commissioned in several countries in the last few years have demonstrated the feasibility of providing highly reliable communication and other advantages over conventional systems despite the early stage of development. The optical fiber system is already cost competitive with copper-based systems for applications carrying 480 (34 M bits) or higher bit rates per fiber. With the reduction in costs of pro- duction of fiber in a few years because of improvements in technology and increased volume, the optical fiber systems are expected to be competitive for 8 M bit rate systems (120 channels) and by the end of the 1980s for the 2 M bit rate systems. A number of countries, including the United States and the United Kingdom, have ordered several systems for regular commercial operation on main routes to be commissioned between 1981 and 1983. Most of the development and application of lightwave communication have been and - 39 - will continue to be for digital transmission though there have been some short-haul analog transmission systems for video (television) links. G17. The capacity of a digital lightwave system is the maximum rate at which the pulses can be sent and received; this rate is limited by how much the signal is distorted by the dispersion (spread over time) of the pulses as they travel along the line. The greater the dispersion, the longer must be the time slot for the receiver to sense the presence or absence of pulses accurately. Dispersion in the optical fiber is caused either by modal dis- persion due to spreading of light along different paths or by modes of dif- ferent lengths in the fiber with different travel times; also the longer the fiber, the more is the dispersion. The highest capacity fibers have only a single mode with no modal dispersion; however, such fibers are much smaller, more difficult to couple light into, and harder to splice and connect than other types. The fibers that are first being applied commercially are multimode of the stepped-index or the graded-index types--mostly the lat- ter. The dispersion in these fibers is reduced by grading the index of refraction across the fiber core--highest at the center and progressively decreasing away from the center. With such a construction, light following shorter paths travels more slowly--and vice versa--so travel times tend to equalize. A graded-index fiber transmits more than 1,000 M bits per second for 1 km in the better present day systems. G18. Signal distortion is also a function of the wavelength range (purity of frequency) of the light pulse. Each transmitter emits light at a certain wavelength such as 0.8 micrometer. If the light is not pure and contains other wavelengths, these travel at different speeds causing pulse dispersion as it traverses a length of the fiber. This is called chromatic dispersion because it depends on the colors of the light and also on certain characteristics of the glass of which the fiber is made. G19. Light emitting diodes produce a relatively broad range of wave- lengths and at the 0.8 micrometer wavelength, range is limited in current systems to about 140 M bits/sec for 1 km path. Semiconductor lasers emit much purer light and about 2,500 M bits/sec may be transmitted over 1 km path. G20. Lightwave systems now finding application use graded-index fibers at a wavelength of about 0.84 micrometer--in the near infrared portion just below visible red. Current systems provide for transmission of 34 M bits/sec with spacing of up to 8 km. A fiber costs more than a wire pair but less than a coaxial tube. The cost per voice circuit-kilometer dimin- ishes as the number of circuits carried increases. At low capacities (cur- rently below, say, 120-480 channels), lightwave applications are limited by economics, at high capacity (above, say, 1,920 channels) by technology. G21. A variety of cable designs have been successfully implemented to meet different service requirements but there- is no standardization as yet-- this will change as cost effective designs become more available. - 40 - G22. Reliable aluminum-gallium-arsenide devices, including fast LEDs and linear lasers for the 0.8 micrometer wavelength are available from sev- eral sources. Avalanche photo diodes (APDs) are being used extensively. Silicon pin photo diodes have also been used successfully with FETs to achieve low noise receivers. G23. The minimum fiber loss/km at 0.8 micrometer is about 3 db; at 0.9 micrometer, 2 db; at 1.3 micrometers 0.5 db; and at 1.5 micrometers even lower. The 1.3 micrometer range will permit larger repeater spacing and is additionally attractive because, in silica-based fibers, modal dispersion is minimal at this wavelength. New systems components are now being actively pursued and are expected to find application in about a decade or so. At that time, each fiber can be expected to operate on a multiplex basis in different bands, several broad-band systems providing for system expandabil- ity and lower costs. Choice-between-24-channel and 30-channel PCM Systems G24. Digital exchanges are designed to work integrally compatible with either the CEPT 30-channel or the American 24-channel type PCM systems (paragraphs F6 and F7). The choice between either type of PCM system is fundamental, with major long-term implications. The choice has to be made carefully because once made, it would effectively rule out any future use of digital switching and transmission systems compatible with the other type. While both types are capable of providing most if not all of the features of digital systems, the CEPT compatible type digital systems have some advan- tages over the American compatible type digital systems for two reasons. G25. The CEPT PCM system provides one 64 k bit channel for signaling and a second channel for time frame synchronization, etc. The signaling channel can be arranged flexibly to provide combinations of signaling and data transmission as are required while fully retaining code (bit) transpar- ency. On the other hand, the 'bit stealing' method of signaling in American type systems destroys such code transparency and may inhibit the development of some mixed data and speech transmission network and of integrated ser- vices data network (ISDN)--the latter defined as an integrated switching and transmission network in which the same digital switches and digital paths are used to establish connections for different services, e.g., speech, data, telex, facsimile, teletext, view data, etc. G26. A second reason for preferring CEPT (30-channel) type is the availability of switching equipment from varied sources. The USA, Canadian and Japanese suppliers manufacture digital exchanges compatible with 24-channel PCM equipment, though digital exchanges compatible with 30-channel PCM equipment are also available from the Japanese and some North American suppliers at this time. Also, the number of manufacturers from these countries offering 30-channel compatible digital exchanges is likely to increase in the near future. On the other hand, the European suppliers manufacture digital exchanges compatible with only 30-channel PCM equipment, and with a few exceptions are likely to continue to do so in this decade. - 41 - Hence, a decision to adopt only 24-channel PCM working would bar European suppliers from bidding for digital exchanges, whereas a decision to adopt the 30-channel digital exchanges would be considerably less restrictive in bidding and provide more competition. Manufacturer Support Needs G27. Manufacturers' support is essential with any switching system. With conventional exchanges, apart from the intitial assistance for instal- lation and training of staff, support services through the life of the equipment are required to sort out bugs if any, and supply of spares and supplies of additional equipment to expand to full capacity. Adjustments and repairs to equipment are done locally with simple tools and guages. With digital exchanges, the degree of support needed from manufacturers includes all of the above and more, particularly for software. G28. Digital exchange hardware components are highly reliable and the number of faults are likely to be few. When printed cards go faulty, these are replaced and the faulty cards tested and repaired in a central shop or by the manufacturer. While the repair and testing of cards or components are relatively straightforward, testing requires the use of minicomputers with special programs for each type of card or component. While there are no technical difficulties in repairing and testing of these cards in devel- oping countries, the required facilities for testing all types of cards may not be justified in small developing countries with only a few exchanges. Under such conditions, the faulty cards may have to be sent to the manufac- turer for repair and return. G29. Manufacturers' support for software will be required, except in a few advanced developing countries, for some or all of the following: (a) initial supply, loading and commissioning of software, including local assistance for 3 to 12 months after com- missioning for debugging initial faults; (b) updating of generic software from time to time for adding new facilities, incorporating improvements, or remedying defects which come to light subsequent to the initial debugging--these defects become known to the manufacturer on a global basis so that these updates are useful in making improvements even before faults show up locally; (c) adaptation of and interfacing with new signaling systems; (d) changes to provide new data for technical or operational administration; and (e) updating data base (site application software) to the extent these are not done by local staff. - 42 - Procurement G30. In general, the requirements for digital switching systems are similar to those for conventional systems. The preparation of bidding docu- ments and evaluation of received bids are also similar. The requirements to be met, the preparation of bid documents and evaluation of bids for all switching systems including digital SPC systems have been dealt with compre- hensively in the new CCITT GAS 6 Handbook, "Economic and Technical Aspects of the Choice of Switching Systems." This handbook was finalized in June 1980 and is expected to be available in book form from ITU in early 1981. This handbook will be of great help to telephone administrations in drafting technical specifications, and other procurement related work for switching systems. Other useful CCITT publications are also likely to become avail- able in 1981 (paragraph G32). G31. Apart from the general requirements, administrations deciding on procurement of digital switching systems at this time will have to decide on the following: (a) decide between 30-channel and 24-channel compatible digital systems (paragraphs E20 to E22); (b) decide on the extent of manfacturer support for hardware and software (paragraphs E23 to E25) and the training requirements needed for inclusion in the bid invitation documents; and (c) decide on whether the bids should be confined to fully proven systems only, or bids of other systems not proven would also be acceptable. If the decision is for the latter, clauses for qualifying bidders should be included to reduce the risks, that is to ensure that they have the necessary experience and competence, a record of provid- ing manufacturer support in developing countries though not particularly for digital systems, that the system offered, or at least each of its main subassemblies, has been tried out in at least one public service installa- tion, and that the manufacturer has a clear commitment to regular production of the offered system. The Bank has assisted borrowers to draft such clauses in recent pro- curement of digital systems and these can be referred to, if needed. Forthcoming CCITT Publications on Digital Systems G32. The CCITT has drawn up a series of important recommendations and useful manuals dealing with digital switching andtransmission systems, including common channel signaling system 7 and program languages for SPC systems. Publications approved in the November 1980 plenary session of CCITT should be available in 1981. This paper is based on information gathered from many publications of which the principal ones are listed at Annex 3. - 43 - ANNEX 1 ANALOG AND'DIGITAL SIGNALS 1. The movement of intelligence from one point to another is the basic task of telecommunications. The intelligence to be moved can be called a message, regardless of the form it takes or its purpose. The most common form of message conveyed in telecommunications systems is speech and the telephone system was initially developed for voice communications. Over the years, however, many other types of messages have evolved, e.g., facsim- ile, radio program, video and data. Telecommunications technology first translates these messages into electrical signals and provides communication channels which transmit the electrical signals from source to destination where the electrical signals are converted back to the original message. 2. The variety of signals and types of communication channels inter- act in many ways requiring channels of various bandwidths and operating characteristics. The signals can be in analog or digital form and can be converted from one to the other, if necessary. Also, transmission from source to destination can be in analog or digital form. 3. The essential common feature of analog and digital signals is the presentation of electrical signals which are representative of the measured levels of physical features of the message. The mode of presentation between analog and digital can be explained by comparing them with graphs and tables to present measured data. As the graph behaves on the paper in the same way as the measured physical parameter does in the physical pro- cess, the graph is said to be an analog representation of the physical parameter. A table reproduces the behavior of the physical parameter by means of a number of measurement values expressed by digits, i.e., a table is a digital representation of the parameter. Conversions between graphs and tables are of course possible as they represeent the same information. Going from a graph to a table, we choose a number of points on the graph and read the numerical values from the scales on the axes. In the other direc- tion from table to graph, we plot the table values and draw a smooth curve through them. The scales and number of points must be chosen that the graph is good enough for the particular purpose--an exact replica is neither pos- sible nor necessary. 4. The most common signal transmitted in telecommunications is the speech signal, an electrical signal generated in the telephone set as an analog of acoustical speech wave generated in the voice box or larynx of the speaker, viz. the variations of level over time of the electrical signals correspond to the variations of sound pressure over time of the speech. In pulse code modulation of speech, the speech is conveyed as in a table, where the table values are coded in electrical form. The pulse code modulator - 44 - chooses a number of points on the analog electrical speech signals, measures their numerical values from the scales which represent amplitude and time, and transmits the numerical values to the pulse code demodulator at the destination. The pulse code demodulator plots the speech table values and draws a smooth curve between the points, i.e., reconstructs the analog speech signal. 5. Obviously a choice of the scales and the number of points has to be made so that the reconstructed signal is good enough for reproducing the original speech (or other signal) adequately. For economical reasons the number of table values and accuracy has to be limited. PCM techniques opti- mizes these parameters to suit the transmission channels at least cost while providing the needed quality. The principles and methods used in such opti- mization are not dealt with here but is treated very well in textbooks and articles; a recommended reading for this purpose particularly for nontechni- cal persons is at reference 9 listed in Annex 3. 6. The basic telegraph, teletext or data signal consists of a train of pulses which represent in coded form the information to be transmitted. Such signals are created and processed in many ways to make them suitable for transmission from origin to destination. Generally telegraph and data signals are digital (but not PCM) in nature. The pulses can be of different amplitudes, duration or have different relative positions. 7. Picture, facsimile and other transmission used analog signals until recently. However, recent developments in coding (not PCM) has enabled the information for transmission of pictures and facsimile to use digital signals efficiently and economically. - 45 - ANNEX 2 24-CHANNEL AND 30-CHANNEL PCM SYSTEMS 1. CCITT has recommended two first order PCM systems for use in the telephone network-the 30-channel system proposed by CEPT and the 24-channel system of the AT&T in USA (paragraphs F6 and F7). The first order or primary system in both types form the basis for hierarchical multiplexing of PCM systems to provide multiples of the number of circuits in the first order systems. 2. Both systems sample the speech signals at 8,000 times per second and 8 binary bits are used to represent the value of the sample. However, the number of channels and the coding and signaling arrangements are quite dif- ferent. The descriptions of the two systems are given in CCITT recommenda- tions and in several publications. The principal characteristics of the two first order systems are given below: 30-Channel System 24-Channel System Audio frequency band 300-3,400 Hz 300-3,400 Hz Sampling rate 8,000 Hz 8,000 Hz Bits/sample 8 8 1/ Time slots/frame 32 24 PCM Channels/frame 30 24 Output bit rate 2,048 kbit/s 1,544 kbit/s Encoding law A-law,A=87.6 u-law, u=255 Signaling capacity Channel associated signaling 1-4 sign. ch./PCM ch. 1-2 sign. ch./PCM ch. Common channel signaling 64 kbit/s 4 kbit/s 1/ Only 7 bits each 6th frame if channel associated signaling is used. 3. The hierarchical multiplexing arrangements are as follows: 30-channel Type Hierarchical level 1 2 3 4 - 5 Number of channels 30 120 480 1,920 7,680 Bit rates in Mbits/second 2.048 8.448 34 140 560 24-channel Type Hierarchical level 1 2 3 4 Number of channels 24 96 672 4,032 Bitsrates in Mbits/second 1.544 6.132 44.7 274.2 - 46 - ANNEX 3 REFERENCES Title Author Source 1. Economic and Technical Aspects of the Choice GAS 6 of Switching Systems Draft Handbook, ITU, Geneva 1/ 2. Introduction to Exchanges with Lacout TELECOMMUNICATION JOURNAL, Stored Program Control Vol. 46-IV, 1979 3. Choice of Switching Systems B. H. Shanta Pai TELECOMMUNICATION JOURNAL, Vol. 46-IV, 1979 4. Beginnings: Global Transition to Digital G. A. Langley TELEPHONY, July 10, 1978 Switching is Under Way 5. The Impact of New Technologies G. A. Langley TELEPHONY, June 18, 1979 6. Economic Aspects of Introducing Digital G. Doyon INTELCOM, 1979 Transmission and Time Switching in Urban Dallas, Proceedings and Rural Telephone Networks 7. Project ESS - Updates VI, VII and IX Dittberner Associates, Inc. 8. Understanding Microprocessors Texas Instruments Learning Center, Published for Radio Shack 9. Digital Telephony: An Introduction L. M. Ericsson 10. Introduction of Digital Telephone Exchanges P. Nolke SIEMENS TELECOM REPORT in Analog Networks. January 1980 11. Continental Digital Experience B. M. Barbie TELEPHONY and others June 23, 1980 12. Long-term Technical Plan of a Country T. Larson and TELE, January 1980 Advanced in Telecommunications L. Ackyell 13. Operation and Maintenance of Networks with AXE 10 ERICSSON REVIEW, March 1979 14. AOM 101: An Operation and Mainterkance System ERICSSON REVIEW, September 1979 15. Digitizing Trunk Network in France J. Verree COMMUNICATION AND TRANSMISSION March 1980 16. Radio Link Equipment-for Digital P. Dallot COMMUNICATION AND TRANSMISSION 2 x 34 MBit Transmission and others March 1980 17. Lightwave Systems - An Overview Snyder and TELECOMMUNICATION JOURNAL, Cohen Vol. 47-VI, 1980 18. Fiber Optics C. Kao TELEPHONE ENGINEER AND MANAGE- MENT, September 1, 1980 19. Lightwave - Yesterday, Today and Tomorrow Jacobs TELEPHONE ENGINEER AND MANGAGE- MENT, September 1, 1980 20. Optical Fibers and Systems C. Kao ERICSSON REVIEW, March 1979 21. The Fiber Future L. C. Gunderson TELEPHONY, September 29, 1980 1/ Expected to be published in 1981.